This invention relates to a cut-through transmission apparatus and a cut-through transmission method for use in a node included in a communication network.
In recent years, the internet traffic is rapidly increased. The internet traffic is carried by IP (Internet Protocol) packets which are transferred under control of a router. Following the rapid increase in internet traffic, there is a growing demand for improvement in function and performance of the router. Since most of transferred data are important data for business use, a strict demand is imposed upon the reliability, the quality, and the security. On the other hand, there arises an increasing demand for a virtual private network, such as an internet VPN service, connecting a plurality of sites or nodes.
In order to meet the above-mentioned demands, one approach is to make the router have a high performance and a full of additional functions. However, this approach has a limit. There remains a problem how to realize the above-mentioned demands in an IP packet network in association with an existing transmission network.
As the existing transmission network, there are known an SDH (Synchronous Digital Hierarchy) network, an ATM (Asynchronous Transmission Mode) network, a WDM (Wavelength Division Multiplexing) network, and PDH (Plesiochronous Digital Hierarchy) network. By integrating the existing transmission network and the IP packet network transmitting the internet traffic and by complementing their characteristics with each other, it is expected to achieve an improved system having more efficient transmission characteristics.
Consideration will be made about conventional network structures for LAN-to-LAN connection.
Referring to FIG. 1A, four nodes A through D are connected through a plurality of SDH paths 100 in a point-to-point connection. Specifically, the SDH paths 100 as private lines are provided between the nodes A and B, between the nodes A and C, between the nodes A and D, between the nodes B and C, between the nodes B and D, and between the nodes C and D, respectively. This network structure called a mesh type is disadvantageous in that a large number of SDH paths (private lines) are required.
Referring to FIG. 1B, four nodes E, F, G, and H are connected through an IP packet network of a hop-by-hop connection. Specifically, a plurality of SDH paths 101 as private lines are provided between the nodes E and F, between the nodes F and G, between the nodes G and H, and between the nodes H and E. In addition, each of the nodes E, F, G, and H has a router function to execute a routing operation.
Referring to FIG. 2, a router 102 performs the routing operation as a third-layer operation. Specifically, the router 102 individually identifies all packets flowing into the router 102 to judge whether each packet is to be dropped at this node or to be hopped to a next node.
In order to execute the routing operation, the router 102 is required to have a number of functions. Thus, a heavy load is imposed upon the router 102.
Specifically, the router 102 is required to have a large routing table including routing information for all packets flowing into the router 102 and to have a high performance so as to search such a large routing table.
In addition, in order to interface the SDH paths (the private lines) in the form of DS-1 (Digital Signal Level 1), the router 102 is required to have a path terminating function of terminating VT (Virtual Tributary level) 1.5.
Since all of the packets are sent to a third layer after being subjected to the above-mentioned operations, a processing delay is produced also for those packets which are to be hopped to the next node and need not be processed at this node. This results in significant degradation of a network performance as a whole.
Thus, packet transfer in the above-mentioned IP packet network is disadvantageous in that the third-layer operation, i.e., the routing operation inevitably gives the heavy load.
For example, Japanese Unexamined Patent Publication (JP-A) No. H10-136016 (136016/1998) discloses a packet transfer control method which is capable of shortening a time required for a router to create a private cut-through path for a particular end flow.
The operation of the above-mentioned packet transfer control method is as follows. It is assumed that, in order to transfer a packet flow defined by a relatively abstract (general) condition such as a destination network address, a first cut-through path is preliminarily established from a first router to a second router which is not adjacent to the first router. In this state, it is assumed that a second cut-through path is required to be established to transfer a packet flow defined by a more specific (detailed) condition such as a source/destination address pair and a destination port number. In this event, control messages for establishment of the second cut-through path are exchanged between the first router and the second router as an endpoint of the first cut-through path. Thus, the second cut-through path is established by the use of the first cut-through path.
In the above-mentioned packet transfer control method, cut-through connection is established by the router. As a result, an operation load imposed upon the router is not reduced.
On the other hand, Japanese Unexamined Patent Publication (JP-A) No. H10-294737 (294737/1998) proposes a packet transfer apparatus which is capable of reducing a delay which is produced upon start of transfer of a best-effort flow and required to establish a connection between the packet transfer apparatus and another packet transfer apparatus.
In the packet transfer apparatus, each router containing a switch monitors a time duration of the flow. If the flow continues for a long time, a cut-through connection is established by switching and assigned to the flow. If the flow continues for a much longer time, a short-cut connection using an ATM connection is established and assigned to the flow. Thus, a basic router is avoided.
In the above-mentioned packet transfer apparatus also, cut-through connection is established by the router, like in Japanese Unexamined Patent Publication (JP-A) No. H10-136016. As a result, the operation load imposed upon the router is not reduced.
Japanese Unexamined Patent Publication (JP-A) No. H09-172457 (172457/1997) proposes a packet transmission node apparatus which is capable of establishing cut-through connection selectively for a traffic expected to have a relatively large amount of communication after establishment of the cut-through connection.
In the packet transmission node apparatus, a node which can be a starting point or an end point of the cut-through connection refers to, before transmission or after reception of a packet, not only information of a network layer of the packet but also at least one of source information and destination information of a transport layer. If it is judged as a result of the reference that establishment of the cut-through connection is worthwhile, initiation of connection establishment is triggered by the packet.
In the packet transmission node apparatus, the cut-through connection is established not in the router but in the node. Therefore, the load upon the router is reduced. However, since the various kinds of information must be referred to as mentioned above, packet transmission inevitably requires a complicated operation.